In this talk, we will present a new approach for approximating
large independent sets when the input graph is a one-sided
expander—that is, the uniform random walk matrix of the graph has
second eigenvalue bounded away from 1. Specifically, we show...